Low Poly vs High Poly: How to Choose the Right Polygon Count
Why Polygon Count Matters
Every 3D model is built from polygons, usually triangles or quads. The number of polygons determines how much detail a mesh can show, but also how much work the computer has to do to display it.
A character with 2 million polygons can look stunning in a still render and bring a mobile game to its knees. A 5,000-polygon version of the same character can run smoothly on a phone but looks blocky in a close-up. Neither is "better": they are built for different jobs.
What Counts as Low Poly and High Poly?
There is no official threshold, and the numbers change as hardware improves. As a rough guide for a single hero asset:
- Low poly: a few hundred to around 20,000 triangles
- Mid poly: roughly 20,000 to 100,000 triangles
- High poly: from about 100,000 triangles up to several million
Context matters more than the number. A 10,000-triangle rock is heavy for a mobile game, while a 60,000-triangle character is normal for a modern console title.
Low Poly Models
Best for
- Mobile, web and VR games where performance is critical
- Background props, crowds and large environments
- Stylized art (flat-shaded or "low poly" aesthetic)
- AR experiences and product viewers on the web
Advantages
- Fast rendering and lower memory usage
- Smaller files that load quickly over the network
- Easier to rig, animate and UV-unwrap
Trade-offs
- Fine detail must come from textures and normal maps
- Silhouettes can look angular up close
High Poly Models
Best for
- Product visualization and marketing renders
- Film, VFX and cinematics
- Architectural visualization close-ups
- 3D printing, where real geometry is what gets printed
- Source models for baking normal maps
Advantages
- Smooth curves and real geometric detail
- Look great from any angle and distance
- No need to fake detail with textures
Trade-offs
- Heavy files and slow viewport performance
- Usually unsuitable for real-time use without optimization
- Harder to edit, rig and animate
The Best of Both: Baking Normal Maps
Most professional game assets use both. Artists sculpt a detailed high poly model, build a clean low poly version on top (retopology), and then bake the high poly detail into a normal map.
The normal map tells the renderer how light should hit each pixel, so the low poly model appears to have wrinkles, bolts and seams that are not actually in the geometry. The result looks close to the high poly original at a fraction of the cost.
Common tools for this workflow include Blender, Substance 3D Painter, Marmoset Toolbag and ZBrush.
How to Reduce Polygon Count
If you already have a high poly model and need a lighter version, you have several options:
- Decimate modifier (Blender): quickly reduces triangle count; great for static props and scans.
- Retopology: rebuild a clean mesh by hand or with tools like Quad Remesher or Blender's remesh options; best for characters that will be animated.
- LODs (levels of detail): create several versions of the same model and let the engine switch between them based on distance. Unity and Unreal both support LOD groups.
- Engine-level tools: Unreal Engine's Nanite can render very dense static meshes efficiently, which changes the rules for some projects, but it does not apply to every platform or to skinned meshes in every version.
Always check the result in your target engine; a model that looks fine in the viewport can still show shading artifacts after heavy decimation.
Quick Decision Guide
- Mobile or web game? Low poly with baked normal maps.
- PC or console game? Mid poly hero assets, low poly props, plus LODs.
- Still renders or animation? High poly, or mid poly with subdivision at render time.
- AR or web product viewer? Low to mid poly GLB with PBR textures.
- 3D printing? High poly, watertight geometry.
Find the Right Model for Your Project
Every product on todo3ds lists its polygon and vertex count in the specifications, so you can check whether a model fits your performance budget before you buy.